This is most likely negligence, but I typically have 3-4 VMs open. One is for hotfixes, one for next feature branch, one for secondary project, one for another experiment. It keeps things really clean from a deploy point of view and reduces the cost of context switching greatly. This is VMware stuff running on Windows 7. I'd love to hear more ideas. All 16GB of my random is always in use.

Doesn't have to be development related try editing 4K video with 16GB of ram.

Agreed but that's not necessarily bad on 16GB. After all, you can edit 4k on an ipad pro with 2-4GB ram. Disk IO matters a lot though - newest MBPs do 3.1GB/s so having a bit less RAM is nothing like it used to be before fast SSDs.

No you can't edit a video on an iPad pro at 4K, adding a few clips together isn't editing.

Open Adobe Premiere and After Effects load up a few 4K sources and start editing and talk to me about 2-4GB ram for 4K....

Splicing clips and adding transitions definitely is editing. What you're talking about is differences in workload, which of course have different IO and CPU requirements. The further you get into those complex workloads, the fewer users there are, and the more likely they're going to be sitting at a fixed station with a large monitor where an iMac / MacPro may be a better choice for said workload.

Even if MBPs had 32GB, you still will hit CPU & thermal limits which matter a great deal in rendering, and which are far less than an iMac. I imagine you're looking at a pretty limited set of use cases where 32GB freed up a speed bottleneck that the CPU didn't then quickly hit. Even more so factoring in these use cases needing to be mobile.

Also, Premiere / After Effects can most definitely take advantage of the improved disk IO, now that its available. As usual, developers will have to adjust to the platforms they're on. Having graphics performance doubled and disk IO up 66% is not a terrible state of affairs for rendering.

I'm not saying 32GB isn't important but there's a pretty understandable logic as to what fundamental constraints they're dealing with and how those compare to the market.